Introduction
Anthropic's Fable 5 is widely regarded as the most powerful AI model on the market, yet new data suggests that U.S. companies are not rushing to adopt it. According to spending data from financial services provider Ramp, Fable 5 accounts for only about six percent of tokens purchased from Anthropic in its first month after launch. This slow uptake raises important questions about the economics of frontier AI and whether corporate willingness to pay has hit a ceiling.
The Data: Fable 5's Modest Market Share
Ramp's analysis, based on its proprietary token spend management product, reveals that Fable 5 generated only 11.4 percent of Anthropic's model-related spending in its debut month. In contrast, OpenAI's flagship model, GPT-5.6 Sol, captures 25 percent of tokens and 23 percent of spending at OpenAI. Overall, Fable 5 brought in only about 75 percent of the revenue that GPT-5.6 Sol generated, despite costing significantly more per token.
It's important to note that Ramp's sample skews slightly toward tech companies, which are more likely to adopt cutting-edge AI. This suggests that actual Fable 5 adoption across all industries could be even lower. The data indicates that even among tech-forward firms, the premium price of Fable 5 is a significant barrier.
Price as a Barrier
Ramp economist Ara Kharazian attributes Fable 5's slow uptake primarily to its price. The model costs about $10 per million input tokens and $50 per million output tokens, making it roughly twice as expensive as GPT-5.6 Sol or other Anthropic flagship models. Kharazian sees this as evidence of a new ceiling on what companies are willing to spend on AI, arguing that the extra performance simply isn't worth the cost for many use cases.
This price-performance trade-off is a critical consideration for businesses. While Fable 5 may offer superior capabilities, the incremental benefit over cheaper models like GPT-5.6 Sol may not justify the doubling in cost. For many applications, the performance difference is barely measurable in daily work, making it difficult for companies to rationalize the expense.
The ROI Conundrum
The slow adoption of Fable 5 highlights a fundamental challenge in calculating AI return on investment (ROI). How does a company quantify the value an AI model delivers, especially when trying to measure the gap between one model generation and the next? The value of AI is often abstract and difficult to translate into concrete financial terms. This complexity makes it hard for businesses to justify premium prices for marginal performance gains.
Moreover, the data doesn't necessarily indicate that companies have reached an absolute limit on AI spending. Rather, it suggests that they are unwilling to pay for incremental improvements that don't translate into tangible business outcomes. If a model were dramatically more capable, delivering clearly measurable value, companies might be willing to pay more. But as long as the value remains abstract, their willingness to pay appears to be capped.
Implications for AI Providers
For AI providers like Anthropic, this data serves as a cautionary tale. Simply being the most capable model isn't enough to drive adoption if the price isn't aligned with perceived value. Providers may need to reconsider their pricing strategies or focus on demonstrating tangible ROI to enterprise customers. Alternatively, they could develop more cost-effective models that meet the needs of a broader market.
The deceleration in growth at both OpenAI and Anthropic, as noted by Ramp, further underscores the need for sustainable business models. As competition intensifies and customers become more price-sensitive, AI companies must find ways to deliver value that justifies their costs.
Conclusion
Fable 5's slow adoption is a telling indicator of the current state of the AI market. While frontier models continue to push the boundaries of capability, corporate spending is constrained by practical considerations of cost and ROI. The data from Ramp suggests that the willingness to pay for frontier AI has indeed hit a ceiling, at least for now. This doesn't mean the ceiling won't be broken in the future, but it does signal that AI providers must adapt to a more discerning and value-driven customer base.
This article is based on reporting by The Decoder. Read the original article.
Originally published on the-decoder.com
